Practical Considerations for Boosting Models in Algorithmic Trading

Summary

The document raises practical questions about using XGBoost to predict whether the next candle’s closing price will move above or below a chosen threshold from the current open. The proposed feature set combines market sentiment, macroeconomic variables, broad equity index data, and technical indicators for the cryptocurrencies being modeled. The author questions whether reported boosting-model results that outperform buy-and-hold for Bitcoin and Ethereum can be reproduced in live trading.

The text frames concerns about time-series modeling and industry use but supplies no answers, experiments, or trading results of its own. It does not describe a validation scheme, execution assumptions, transaction costs, or safeguards against leakage, so the cited performance claim cannot be assessed from this document. Its value is primarily as a prompt to investigate realistic evaluation and implementation constraints before treating reported backtests as tradable evidence.

Key ideas

• The proposed task is to classify whether a future candle close exceeds a threshold relative to the current open.
• The described features mix sentiment, macroeconomic data, equity index prices, and cryptocurrency indicators.
• The author questions whether published boosting results can translate into practical trading performance.
• The document provides no model evaluation or answers about validation, execution costs, or industry adoption.

Tags

# Boosting models for algo trading


# Boosting models for algo trading












I’m currently working on a xgboost model to predict the price change above or below a given percentage between a candle’s open price and the next candle’s close price. I use a wide range of features, including market sentiment, unemployment rate, inflation, s&p ohlc data, as well as calculated technical indicators for the coins at hand. I have seen a recent article on achieving performance vastly superior to hodl on ETH and BTC with boosting models in particular (https://arxiv.org/pdf/2311.14759.pdf), but I’ve grown skeptical to whether these numbers could be achieved in practice. The question is to the people who have implemented similar methods for work or for their own projects — what are the considerations in terms of using boosting for time series analysis and specifically algo trading? Are boosting models even used in the industry for such tasks? Thanks in advance to whoever sheds some more light on this matter, I’d be very grateful!

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.